International Small Business Journal: 602 International Small Business Journal Entrepreneurship policy to support new innovative ventures: Is it effective

http://isb.sagepub.com/International Small Business Journal

http://isb.sagepub.com/content/28/6/602The online version of this article can be found at:

 DOI: 10.1177/0266242610369874

2010 28: 602International Small Business JournalCharlotte Norrman and Lars Bager-Sjögren

Entrepreneurship policy to support new innovative ventures: Is it effective?  

Published by:

http://www.sagepublications.com

can be found at:International Small Business JournalAdditional services and information for    

  http://isb.sagepub.com/cgi/alertsEmail Alerts:

http://isb.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://isb.sagepub.com/content/28/6/602.refs.htmlCitations:  

What is This? 

- Jan 7, 2011Version of Record >>

at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from at Linkoping University Library on September 23, 2013isb.sagepub.comDownloaded from

Small FirmsisbjArticle

Entrepreneurship policy to support new innovative ventures: Is it effective?

Charlotte NorrmanLinköping University, Sweden

Lars Bager-SjögrenSwedish Agency for Growth Policy Analysis, Sweden

AbstractUsing Swedish data, we investigate whether the effectiveness of an entrepreneurship policy programme can be traced over time among those firms it supports. The results are drawn from a longitudinal matched pair analysis. Hypotheses were tested every year for eight years. The main conclusions are: when bias is considered the public support programme has not generated measurable additionality and the programme has to some extent been able to select firms on a general level; however, among those selected, the scheme has not been able to identify potentially successful firms.

Keywordsentrepreneurship policy, matched pair analysis, public innovation support, public policy

IntroductionInnovative ventures are considered to be important components in the creation of societal growth and wealth. Thus, during recent decades, large amounts of money have been put into governmen-tal entrepreneurship policy programmes (Vedin, 1993; Storey, 1994; Heydebreck et al., 2000; North et al., 2001; Jaffe, 2002; COM, 2005; Lundström and Stevenson, 2005; COM, 2006; Norrman, 2008). According to policy declarations, it would seem that these efforts will continue in the future (COM, 2005; Ministry of Trade and Industry, 2006; OECD, 2006; Edling et al., 2007). However, in order to measure how successful investments in policy are, the costs of the programmes ought to be weighed against their benefits (Vedin, 1993; Lundström and Stevenson, 2002; OECD, 2006). Policymakers seem to be aware of this and the issue of reliable evaluations is emphasized (COM, 2005).

Corresponding author:Charlotte Norrman, Linköping University, Department of Management and Engineering, SE-581 83 Linköping, SwedenE-mail: [email protected]

International Small Business Journal28(6) 602–619

© The Author(s) 2010Reprints and permission: sagepub.

co.uk/journalsPermissions.navDOI: 10.1177/0266242610369874

isb.sagepub.com

Norrman and Bager-Sjögren 603

As far as we have seen, most studies into the evaluation of early stage ventures are based upon questionnaires that is, self-estimations, and they generally present the programmes surveyed as hav-ing a positive impact (Klofsten et al., 1999; Chrisman and McMullan, 2000). This kind of evaluation is labelled ‘monitoring’ by Storey (2000), and the term implies that the studies are based on descrip-tions of schemes and/or the take-up of opinions on the programme, estimated by its recipients. However, to undertake what Storey (2000) defines as ‘evaluation’, the results must relate to the outcome of a counter-factual, which ideally consists of a group of firms similar to the ones treated in all respect except the treatment. Not considering the ‘counter-factual’ in a correct manner will most probably introduce systematic bias in the estimated impacts, that is, exaggerated estimates.

It is also shown that most evaluations undertaken have a rather short-term approach, which implies that all the effects in the chosen indicators of impact may not yet have emerged. This reduces the prospects of learning from previous policy measures (Rush et al., 2004; Lundström and Stevenson, 2005). Moreover, the area of finance, especially if directed to SME firms, has been studied by several researchers (Leleux and Surlemont, 2003; Storey, 2003), while studies particu-larly directed at the very earliest stages of firm development seem to be more rare (Klofsten et al., 1999; Meyer, 2005).

Finally, a need for more knowledge and for the identification of early, consistent, reliable and cheap information, which can serve as basis for evaluations, is recognized by the policymakers (Mosselman et al., 2004). Against this background, our aim is to investigate the long-term impact of an entrepreneurship policy programme directed at innovative ventures in their early stages. This will be done through following the ‘evaluation, step 4–6’ guidelines of Storey (2000) and making an attempt to identify the existence of additionality, that is, to identify whether a public programme really may be said to have rendered any impact on the economic performance of the supported firms. In this way, we hope to contribute to the discussion regarding the measurement of the dynamics over time of early stages public support, and to the debate on the evaluation of public support for such ventures.

Public support and the scheme studiedIn order to reach a stage of sustainable development, firms need accrue both crucial resources, such as a large enough market, an accepted product or service, purchasing customers and a supportive network. Furthermore, the firm has to be able to utilize these resources in a satisfactory way that is, through its motivation, competence and organization (Barney, 1991; Klofsten, 1992). Taking an idea to market is costly and it often requires larger resources than the firm can provide on its own. Difficulties in getting the idea funded by private-sector actors before customer incomes is gener-ated is therefore, one of the most common barriers for small, new and innovative ventures (Penrose, 1959; Drucker, 1985; Storey and Tether, 1998; Lundström and Stevenson, 2002; Oakey, 2003; Norrman, 2008). These difficulties occur since banks can be reluctant to take risks and because new ventures commonly lack a track record and are seldom capable of putting up security, which implies that they do not qualify for bank loans (Rothschild and Stiglitz, 1976; Deakins, 1994; Shane and Cable, 2002; Storey, 2003). Venture capital (VC) might be an option as VCs take calcu-lated risks. However, VCs are highly selective, which implies that they tend either to pick what they identify as winners or find firms that could be developed into winners (Baum and Silverman, 2004). To be selected for VC funding, firms need to be ‘investment ready’ (Mason and Harrisson, 2002). This normally implies a certain level of maturity, and explains why VCs are said to invest in firms that are relatively close to market launch (Bygrave and Timmons, 1992). Other kinds of obstacles are the various types of market and systemic failure (Storey, 1994; Audretsch, 2002;

604 International Small Business Journal 28(6)

Lerner, 2002; Salmenkaita and Salo, 2002), that is, there might be a demand for a solution, but transaction costs, willingness/ability to pay or ability to benefit from the solution in the short run are limited or non-existent. These obstacles, alone or in conjunction, imply that early stage ven-tures are vulnerable (Stinchcombe, 1965; Klofsten, 1992; Lundström and Stevenson, 2005). At the same time however, they are necessary; this forms the rationale for public support, which forms the rationale for public support. Public intervention in early stage ventures can therefore, act as a tool to reduce the above mentioned obstacles and as a bridge to private sector investments and customer incomes (Oakey, 2003; Norrman, 2008).

The programmeThis study is based upon a financial support programme, the Swedish Innovation Center (SIC) that provided support to ‘innovators in their absolute earliest phases of development with financial capi-tal, advice and networks’ in the years between 1994 and 2003. We consider SIC as an actor that was rather typical of its kind. Its main objective was ‘to create a better innovation climate in Sweden … where people’s attitudes to innovators is positive. And where it is easy for an innovator to receive help to develop his or her concept to a commercialized product or service’ (SIC, 2002: 24).

The SIC programme aimed at a broad mix of businesses in order to create a more competitive edge, and supported the development of ‘practical consumer products as well as advanced tech-niques for industrial and societal purposes’ (SIC, 2004: 2). The financing provided by SIC was seed finance, given in order to verify and develop the ventures further, and it was allocated to both firms and private individuals (Norrman, 2005). To reach the firm owners, SIC set up regional inno-vation centres (RIC) that were both co-located and run in cooperation with the regional ALMI-offices.1 SIC also cooperated with the science parks through their organization, Swedepark.2 At its inception, the SIC raised funds of €56 million (Governmental Proposition, 1993/1994; Governmental Decision, 1994).3 During the programme period, €117 million were allocated, since the fund was expanded through returns from capital investments. The programme was wound up as planned, and transferred to ALMI in 2004.

Two main types of financial support were supplied. The first type, innovation subsidy, was a financial grant of approximately €4000, that was regionally administrated by the RICs. This was normally the first step in what can be described as the SIC funnel process. Promising ideas were directed towards a centrally administrated conditional loan which reached a maximum of €43,500 (observed in this study). ‘Conditional’ refers to the conditions of the loan, as it only had to be repaid if the project generated revenues. Ideas that were judged to be especially promising were encouraged to file multiple applications, and were then funded stepwise. According to the former CEO, this SIC funnel process explains the high rates of support on the loan applications.

Over the years, SIC received 5839 applications for conditional loans. To receive support, a project or idea had to fulfil three conditions. It had to be: (1) new (firms were not allowed to be older than three years);4 (2) be in a position to commercialize; and (3) technically or intellectually advanced (SIC, 2002, 2004, and interviews). The SIC decision process was short, on average 38 days, and support decisions were taken in running order. Fifty per cent of the money was then paid out immediately, 40% after a partial report, and the remaining 10% after the company had filed a final report. Approximately two-thirds of the funding was set aside for measures related to the development and protection of products, while the rest was allocated to support commercialization, marketing measures and other activities, such as negotiation (SIC, 2004).5

To summarize, the SIC programme focused on technology-based (as defined by Klofsten [1992]) ideas and projects. It must be admitted however, that not all of the firms within this study

Norrman and Bager-Sjögren 605

can be labelled ‘high technology’, since SIC also sponsored ‘low’ tech business ventures. However, the requirements of technological advancement and novelty that were set up by SIC (2004) allow us to assume that most of the firms studied can be considered to be engaged with new and innova-tive products or services, at least according to the widest Schumpeterian definition of ‘carrying out new combinations’.

Theoretical start points, workable indicators and hypothesesFor a venture to be sustainable in the long run it has, as argued above, to overcome its initial vulnerability, that is, it has to be able to secure its supply of crucial resources, and it has to be able to utilize these resources in a satisfactory manner. However, when filing an application for seed finance, most ventures are at an earlier stage of development that is, they reside in what Klofsten (2005) labels a pre-commercialization stage. During this stage, the idea is un-formed, difficult to communicate and often difficult for stakeholders, such as potential customers, part-ners and financiers, to understand. When reaching the commercialization stage, the idea is devel-oped enough to be launched on the market and to generate income for the firm and sufficiently mature to attract investment from actors such as VCs. The SIC finance was to be used to test the business idea in different ways for example, measures such as protection (for example, patent), product/prototype development, market analysis and so forth, all measures used to define and clarify the (business) idea.

Linking the aims of SIC to the model by Klofsten (2005) and to the theories of resource depen-dency (Barney, 1991; Klofsten, 1992), leads us to propose that policy programmes directed at early phases of development ought to be concerned with ventures that reside in the pre-commercialization or commercialization stages. Hence, their aim ought to be to develop the supported ventures out of the immature stages and put them on the road towards commercialization, sustainable development and growth. Furthermore, it is our belief that such development ought to be visible. Thus, to reach any conclusion about what impact public interventions may have on this development process, we need to define suitable indicators of performance. Finally, these indicators ought to be strongly cor-related to changes in quality with regard to factors crucial for firm development.

In evaluation research, the cost effectiveness of programmes refers to the cost of inputs in rela-tion to the outcomes generated. In general, outcomes can be related either to stated targets, or to the measurement of impacts which otherwise would not have emerged (Vedung, 1998; Storey, 2000; Nagel, 2002; Mosselman et al., 2004). Although SIC is a public programme, the official publica-tions of SIC6 have not revealed any clearly described ‘programme theory’7 that includes defined targets against which to measure. Turnover (sales) and number of employees were both mentioned as desirable outcomes (SIC, 2004), albeit no explicit evaluation indicators were formulated. This lack of targets blocks estimates of the impact in the case of goal attainment.

Storey (2000) discusses the shortcomings of focusing solely on the outcomes of supported firms as a measure of programme impact. Instead, he argues, impact must be defined as the difference compared to some kind of norm or counter-factual. Thus, in analysing programme impact or addi-tionality, we should focus on the outcomes of relevant indicators that is those which, within a given time limit, would not have been realized without the programme. Impact hence requires a differ-ence in performance between supported and non-supported firms which is due to the support of the programme that is, a comparison of treated versus non-treated. Out of this, a general hypothesis can be formulated.

H1: Supported firms perform better than non-supported firms.

606 International Small Business Journal 28(6)

The next issue was how to measure firm performance. Traditional business ratios (such as solvency and profit margins) might not be the most adequate tool for assessment, since these indicators are linked to the strategic decision making of management and/or investors. Emerging firms are in an early stage of development, that is, they are not yet stable firms, which suggests that these measures can be misleading. In this case, basic and simple measures of economic activity, such as survival, generated sales or the expansion of the firms’ total assets, would seem to be more relevant.

The survival rate of firms is important and can be measured in several ways (Audretsch, 2002; Almus, 2004). The consideration of firm closure is one; however this variable is formal and does not always imply failure, but can also be coupled to mergers or acquisitions. Furthermore, dormant firms can still be officially registered as running and this is complicated to detect. The presence of turnover (sales) is another way of measuring survival. This is coupled to the ability to gather resources, which is crucial to the sustainability of the firm (Barney, 1991; Klofsten, 1992). By attracting sales, the firm can be said to have gained customer acceptance of its products hence, this is an indicator that shows that the firm is on its way to reaching a more mature stage. Sales are also coupled to the SIC aim of facilitating commercialization. Thus, this is the means of measurement that has been chosen. We use a dichotomous variable named ‘the commercialization incidence’. This variable is an indicator of realization of the invention applied for and can be used to check whether, in applying a broad-aiming strategy, SIC managed to help supported firms to commercialize to a larger extent than non-supported firms. It also indicates survival by means of firm activity. A sub-hypothesis 1a would be:

H1a: Supported firms show higher commercialization incidence than non-supported firms.

Firm growth is a common variable in studies of firm dynamics (Harhoff and Stahl, 1998; Audretsch, 2002; Almus, 2004). Storey and Tether (1998) report that entrepreneurs feel that their presumptive growth is constrained by the lack of external financing hence, a contribution in the form of a soft loan ought to imply the opposite and if so, effects ought to show. When studying new ventures, ‘sales growth is a more objective measure than profitability’ (Mainprize and Hindle, 2007: 43). Normally, firm growth is estimated by firm turnover (Chrisman and McMullan, 2000; Davidsson et al., 2002; Almus, 2004). However, with regard to newly established firms sales may fluctuate heavily during the first years; hence, the use of annual levels may result in unfair treat-ment of firms with economic fluctuations. In order to avoid this, we calculate the mean of the turnover (sales) over the years measured. Translated into a further sub-hypothesis it reads:

H1b: Supported firms show larger average sales than non-supported firms.8

Growth may also be estimated by the total size of the business that is, the sum of debts and equity. An increase in total assets is a consequence of a venture development as such an increase shows increased ability to gain resources. This measure increases whether the firms investigated have invested returns in product or service development, or have taken on loans or investors to finance their development. Additionally, an ability to raise capital from external actors, such as banks and investors, implies increased credibility for the firm (Norrman, 2008). Formulated into a sub-hypothesis it reads:

H1c: Supported firms have on average larger total assets than non-supported firms.

According to Audretsch (2002), employment has had the attention of both policymakers and researchers since the report by David Birch in 1981. Employment is of high societal interest since

Norrman and Bager-Sjögren 607

increased employment generates increased incomes from taxes. For the SIC programme too, employment enhancement was a desired outcome (SIC, 1999). Within research, enhanced employ-ment figures are used as a growth indicator in studies such as those by Chrisman and McMullan (2000), Davidsson and Henrekson (2002) and Almus (2004).

Furthermore, the employees and their competence are a core resource within a firm since this variable constitutes the firm’s ability to control and utilize its resources (Klofsten, 1992). A suc-cessful penetration of a market commonly leads to increased production which implies organic growth of the firm that is, an increased number of employees. The next sub-hypothesis reads:

H1d: Supported firms generate more employment than non-supported firms.

Finally, according to Stam et al. (2007) public policy should be focused upon prospective growth firms and not on those that aim to stay small. As mentioned above, SIC selected certain projects that were regarded as more promising. As shown in Table 1, these selected firms received approximately €14,000 more than those that applied and were supported only once.

This state of affairs gives us an opportunity to compare the two groups of supported firms. If it is assumed that SIC was able to select firms with high potential, this ought to be demonstrated by the performance (according to sub-hypotheses a–d) of the multiple supported firms. Formed into hypothesis 2, this reads:

H2: Firms supported twice or more perform better that firms rejected or supported only once.

Time patterns are an important and often neglected issue in the study of innovation dynamics and as Rush et al. (2004), Drucker (1985), Vedin (1993) and Oakey (2003) argue, the effects of innovation need time to emerge. There does not seem to be any standard approach with regard to proper time horizons for programme evaluation however, Reitberger (1983) for example, argues that pay-back streams ought to be created no longer than five years after programme start other-wise, the project must be considered a failure. According to SIC’s internal studies, a majority of the supported projects were carried through during a time span of 3.5 years (Pleiborn, 2002). In this study, we measure the hypothesized indicators annually up to the eighth year after application was filed, in order to track the dynamics of the impact of the SIC programme.

Data and estimation methodThis study is based on a matched pair analysis. To conduct such an analysis, similarity of back-ground variables within the material studied must be controlled. Therefore, data qualification is required before analysis is conducted. The cases analysed were chosen as follows. First, we have focused on the applications from limited companies, since these can be supplemented with

Table 1. Mean Estimates of Granted Loans to Limited Companies, €

Group n M SE (min) 1% (max) 99%

Single appl./support. 378 30,049 13,774 1,087 81,548Multiple appl./support 132 44,670 22,250 3,262 127,759

Notes: n = number of observations, M = average amount of support, SE = standard error, min = lowest amount of support, max = highest amount of support.

608 International Small Business Journal 28(6)

accounting data (that is, 2577 applications made by approximately 1800 individual firms). Second, to qualify for the analysis, the quality of the data must be sufficient. Cases of insufficient quality (missing values) and with impossible values (for example, negative equity capital) have therefore, been removed from the data. Third, to increase the possibility that the firm’s development derived from the project applied for, we have removed all firms that reported economic activity over the three years prior to the year of application.

To be able to estimate treatment effects, we needed to identify a relevant control group, since all statements relating to the relative merit of a programme depend upon the availability and relevance of such a counter-factual, to which the impact can be related. In this study, firms whose applica-tions were rejected formed the counter-factual. However, rejections due to administrative reasons and rejections resulting from those who did not follow the rules for application were omitted.

This leaves us with a final selection of 603 applications. Of these, 378 were supported once, 132 two or more times and 93 were rejected. The applications from these firms were filed between 1994 and 2003. In order to obtain a sufficient number of firms to analyse irrespective of the year of application, we have deflated the information in the annual reports using a price deflator from Statistics Sweden. Analysis was performed on all firms. For the sub-hypotheses H1b and H1c, logarithmic figures are used in order to reduce variance and achieve more stable results. This is a standard procedure when distributions are affected by skewness and high kurtosis. The application year is defined as zero (initial year) and the year after this is the first year of follow-up (regarding sales, see table in note 12).

Sufficient data was available up until 2006, which implies that we have data to allow for a follow-up of eight years, depending on the year of application. However, for each follow-up year, the number of valid observations decreases. This is due to two reasons first, firms disappear due to business failure or merger and second, for administrative reasons that is, a seven-year follow-up cannot be made on firms that applied in 2003. This allows a time series for the first three years without admin-istrative dropout. Since this article focuses on longitudinal programme impact dynamics, we do not recommend that conclusions about firm dynamics be drawn from year four onwards.

One limitation of the data is the relatively few observations in the quasi-control group, that is, the firms rejected. The number of rejected firms is 93 while the number of those supported is in total 510, which gives the impression of a liberal selection policy. This might not be the case, but can be explained by the SIC funnel process described above, as ventures that after contacting SIC officials realized that the work involved in an application would not pay back probably refrained from applying. This implies that the group of rejections consists of a mixture of rent-seekers and valid applications. Using register information for the reason for rejections, we have been able to remove projects that obviously do not belong to the population that was eligible for SIC.

Finally, we cannot exclude the fact that a number of our rejections may have applied for support from other public organizations which would imply that this study has underestimated the SIC impact. However, other available support programmes have focused on later stages in venture development, and according to Bager-Sjögren and Lööf (2005) the overlap is small. This makes us confident that the probability of rejected firms having received support from elsewhere at the time of application is small.

Sources of systematic errorsHeckman et al. (1997) describe four possible sources of systematic errors when estimating pro-gramme impact: (1) un-compatible definitions of the dependent variable; (2) unequal economic circumstances for the groups of cases observed; (3) un-compatible populations for the observed

Norrman and Bager-Sjögren 609

groups; and (4) the existence of non-observable variables, which govern self-selection into the programme. In this study, errors arising from items 1, 2 and 3 have been eliminated as we use annual report data. In the case of item 3, the firms analysed have applied as limited companies, which make them equal in respect to personal risks. The fourth item is, in principle, dealt with as we analyse firms that were all self-selected for the programme. However, as Heckman and Robb (1985: 162, 183) and Jaffe (2002) discussed, bias due to self-selection comprises of two parts. The first is determined by the individuals (firms) deciding to apply to the programme and the other is coupled to the programme administrators and their skill in selecting which applications to accept. Both components imply that selection into the programme is not random. By having data on both accepted and rejected applications we can reduce the problem of self-selection bias to the second component above.

The presence of administrative bias implies a problem isolating whether the effect of the programme is coupled to the treatment per se – financial support in our case – or to the charac-teristics of the firms treated. Presence of an administrative selection bias might imply overesti-mation of the treatment effect of the programme, since the programme officials have been able to ‘pick winners’, which might been successful even without having support. Wallsten (2000) argues for the likelihood of the contrary; that it is reasonable that, in order to secure the results of their programmes, officials tend to prefer ‘safe’ but mediocre firms to firms with radical ideas that are high risk. If the rejected applications contain one successful innovative firm (outlier) this could be enough to produce a (statistical) non-significant programme performance for a given time frame.

Another way of looking at this selection process is that ‘picking’ the most promising applica-tions is a part of the treatment effect. This implies that administrative selection is part of the treat-ment effect that is, a first step in a certification process. We believe that this demands a discussion on how to contrast the policies to make way for ‘good’ companies compared to the policies to foster radical innovative companies.

Since we use firms with rejected applications as a control group in this study, and since we have access to data before applications were made, we are able to explicitly consider both the selection of approved applications as well as the programme itself as a part of the treatment. To estimate the impact of the programme, we apply a matching estimator (Heckman et al., 1997), which helps us to treat the bias. This has the advantage of generating simple means as output; something that is easy to interpret. Another advantage is that no structural assumptions are needed. In principle, the method of matching implies that for every supported firm one firm, or a limited number of firms, is picked within the group of rejected firms. The chosen firms are similar to the supported firms with regard to the set of selected matching variables. By matching in this way, a set of matched pairs consisting of one supported firm and one rejected firm (or a function of rejected firms) is generated. The estimation of impact is performed by calculating the average difference of the cho-sen variables of interests. The use of matching estimators, in different versions, has become increasingly popular in recent years. Matching on several continuous variables has been problem-atic because of the so-called ‘curse of dimensionality’, and most matching estimation has relied on propensity score matching. To counter the weaknesses implicit in earlier methods of estimations concerning cases where the number of observations in the quasi-control group is small (as in our study), Abadie and Imbens (2002) developed a new algorithm, which allows matching on continu-ous variables.9 Thus, the matching completes the construction of the counter-factual or quasi-con-trol group. To compare outcomes we use one-sided hypotheses, which imply that the critical values are different to the conventional two-sided test. The variables used for matching are presented in the next section.10,11

610 International Small Business Journal 28(6)

Table 3. Mean Initial Year, Year 0

Variable Rejected Single sup. Multiple sup.

n M SE n M SE n M SE

Sales incidence 93 0.40 0.49 378 0.38 0.49 132 0.26 0.44Sales (€) 93 30,710 83,341 378 39,834 258,999 132 10,476 32,551Sum of equity (€) 60 27,414 57,950 197 42,862 113,274 64 31,869 83,126Sum of total assets (€) 22 75,740 89,409 83 158,336 313,625 31 83,396 134,462Number of employees 93 0.5 1.7 378 0.6 2.7 132 0.3 0.8

Notes: n = number of observations, M = estimated mean value, SE = standard error.

Table 2. Mean Annual Sales, €

Year Rejected Single sup. Multiple sup.

n M SE Max n M SE Max n M SE Max

1 92 27,633 75,815 413,715 373 55,560 262,764 4,805,915 132 24,459 45,608 290,8283 92 41,519 109,560 796,047 367 89,224 390,154 7,111,014 132 42,317 71,178 560,1735 52 58,421 137,464 707,800 265 127,832 550,031 7,850,386 119 50,504 115,341 1,040,9567 32 70,190 150,006 573,158 161 187,596 766,737 8,296,184 75 59,001 14,8794 1,152,550

Notes: Year = number of years after application was filed, n = number of observations, SE = standard error, M = mean annual sales, Max = turnover of the largest outlier.

Descriptive statistics, findings and discussion

Table 2 presents the descriptive statistics. In the first row, the average sales during the seven years of follow-up are presented. The means displayed give the impression that the group supported one single time is the one that shows the best performance, while the rejected firms are those that show the lowest performance. Even though the group of firms supported a multiple number of times do not excel, we cannot exclude the fact that the performance of the supported groups of firms might be caused by administrative selection. Note the large standard errors and the presence of outliers (max values, showing the largest for each group and year) in all groups, especially within the group of single supported. This is an indication that the most successful firms, in absolute terms, belong to the group of single supported.12

In Table 3, we show the status of the firms in the initial year, that is, the year of application. Differences between supported and rejected firms emerge when the means are studied. This needs to be taken into consideration when the impact is estimated, since these differences may have ren-dered administrative selection bias (Wallsten, 2000). Irrespective of the label, we can conclude that the sum of private equity and sales are larger among the single supported firms. This holds also for the sum of total assets. However, this difference could be caused by the fact that some of the sup-ported firms received their financial support during the year of application, which then implies an increase in their debts.

The mean values displayed in Table 3 exhibit a probable correlation between higher private equity and programme support and according to the results this ought to be an important matching variable. Hence, we have matched firms according to levels of sales and private equity, since these differences were significant in the initial year. The general economic activity can influence certain

Norrman and Bager-Sjögren 611

industries more than others and to consider this, a third and final matching variable, the industrial classification at a two-digit level, is included. The analysis of the variables that constitute hypothesis 1 has covered all the years from 0 (year of application) to eight years after application (Table 4).

To investigate the impact of SIC’s ability to spot winners by use of repeated applications – hypothesis 2 – firms that have applied and gained support once are compared with the firms that have applied and been supported on multiple applications. The same measures as for hypothesis 1 were used. For this analysis too, the cases have been matched as above, with the exception of match according to industrial sector.

Analysing firms without any other matching than the requirement of zero sales in the initial year of application, we find that supported firms experienced significantly higher sales than rejected firms in years, 1–3 (5% level), and years 4–5 (10% level).13 The estimates for the remaining years show the same direction, although this is not significant. Below we show the interpretations of our findings according to hypotheses SH 1a–d and SH 2a–d, starting with the findings from sub-hypotheses 1a and 2a (see table 5).

Commercialization incidenceNo significant estimates were found with respect to the differences between supported and rejected firms (SH1a), and accordingly, the hypothesis cannot be supported. The direction of the estimates is erratic and thus, does not give any indication of impact. When single-time and multiple-time

Table 4. Differences Due to Hypothesis 1a–d: Comparison of all Supported against all Rejected Applicants

n Coef. SE z

Year 1 Sales incidence_1 235 0.060 0.103 0.580 Log of sales mean_1 153 -0.502 0.594 -0.850 Log of total assets_1 72 0.372 0.592 0.630 Number of employees_1 172 0.508 0.437 1.160Year 3 Sales incidence_3 235 -0.065 0.088 -0.740 Log of sales mean_3 188 0.018 0.680 0.030 Log of total assets_3 74 -0.428 0.529 -0.810 Number of employees_3 187 1.107 0.505 2.19**Year 5 Sales incidence_5 164 0.008 0.114 0.070 Log of sales mean_5 115 0.356 0.835 0.043 Log of total assets_5 65 -0.864 0.792 -1.090 Number of employees_5 112 1.217 0.966 1.260Year 7 Sales incidence_7 89 -0.127 0.159 -0.800 Log of sales mean_7 58 -0.133 1.070 -0.120 Log of total assets_7 57 -0.406 1.805 -0.220 Number of employees_7 53 2.564 1.740 1.470*

Notes: n = number of observations, Coef.= coefficient, i.e. the difference measured, SE = Standard error, z = level of significance. Our hypotheses are supported if the critical value is above 1.29 (10% level of significance) respectively 1.64 (at a 5% level). * = 10% level, ** = 5% level.

612 International Small Business Journal 28(6)

supported firms (SH2a) are compared, significant differences in advantage for the multiple sup-ported firms are found, for year 3 (10% level) and for the years five and seven (5 % level). This implies (weak) support for sub-hypothesis 2a. Multiple-time supported firms appear to have a higher commercialization incidence than single application firms.

Average sales The SH1b has to be rejected since the estimates do not reveal any clear direction. With regard to SH2b, the estimates of the average sales are in favour of the single-time supported firms in years one and three and significantly (5% level) in favour in years five and seven. Since the direction is consistent with the opposite direction of the hypothesis, the H2b is (strongly) rejected. A notable difference is that the means for the supported firms are driven to a higher degree by outliers, which might be an indicator of successful innovations, as well as of SICs ability to pick winners.

Average of total assets The findings from H1c show that there are no differences between supported and rejected firms. SH1c is therefore, rejected. For H2c, all estimates are in favour of the single-time supported firms in years one and three and significantly (5% level) in favour in years five and seven. This implies that SH2c is (strongly) rejected.

Table 5. Differences due to Hypothesis 2a–d: Comparison of Multiple against Single Applicants

n Coef. SE z

Year 1 Sales incidence_1 261 0.863 0.071 1.220 Log of sales mean_1 175 -0.219 0.247 -0.880 Log of total assets_1 94 -0.245 0.264 -0.930 Number of employees_1 195 0.019 0.343 0.060Year 3 Sales incidence_3 261 0.095 0.069 1.38* Log of sales mean_3 210 -0.330 0.288 -1.150 Log of total assets_3 94 -0.235 0.340 -0.690 Number of employees_3 212 -0.024 0.435 -0.050Year 5 Sales incidence_5 197 0.227 0.076 2.88** Log of sales mean_5 145 -0.958 0.424 -2.26 Log of total assets_5 84 -1.976 0.693 -2.85 Number of employees_5 141 -0.278 0.625 -0.440Year 7 Sales incidence_7 114 0.192 0.111 1.73** Log of sales mean_7 75 -1.252 0.600 -2.09 Log of total assets_7 75 -3.694 0.976 -3.78 Number of employees_7 72 -1.996 0.8702 -2.29

Notes: n = number of observations, Coef.= coefficient, i.e. the difference measured, SE = Standard error, z = level of significance. Our hypotheses are supported if the critical value is above 1.29 (10% level of significance) respectively 1.64 (at a 5% level).* = 10% level, ** = 5% level.

Norrman and Bager-Sjögren 613

Number of employees

Finally, for the last indicator of performance – the number of employees – the findings of SH1d show a consistent direction with significant differences in year three at a 5% level and in year seven at a 10% level. The direction of the estimates and the significant figures imply a (weak) support for the hypothesis. For SH2d, the picture is less clear, however. The estimates are significantly (5%) in favour of the single-time supported firms in year seven. Taken together, this implies that the SH2d must be considered rejected.

The findings of the main hypotheses 1 and 2, supported firms perform better than rejected firms and multiple supported firms perform better than single supported firms, read as follows. With regard to H1, we show that with the exception of an increased generation of jobs among supported firms there are no measures that show a significant difference between supported and rejected firms which could be coupled to the programme. With regard to the difference between multiple-time and single-time supported firms (H2), the result is less clear. There is a difference in the sales incidence, but the size of sales tells the opposite story. For the remaining two measures, most of the estimates do not support the hypotheses.

How can these results be explained?Table 2 indicates that on average, the administrators made the right choice. Contrary to what has been proposed by others (that is, Davidsson and Henrekson, 2002; Svensson, 2007), on this super-ficial level, governmental support actors seem to be able to identify successful firms. However, we have found a correlation between the amount of private equity (total assets) and support of the application for a loan. When attention is paid to private equity investment in the year of application (that is, by using total assets as a matching variable), the differences disappear (see Table 4). Hence, we can conclude that the decisions to provide support most probably were affected by an administrative selection bias. From a selection point of view, this is highly interesting. It seems reasonable that administrators have regarded entrepreneurs that have risked larger amounts of their own money to be more credible – that is, being of lower risk – than entrepreneurs that have invested less. This behaviour seems reasonable if we recall the facts that (1) the ventures applying for SIC funding were in their earliest stages of development when the applications were filed, which makes them difficult to evaluate, and (2), that SIC spent small resources on evaluating the applications before the decisions to support/reject were taken (see the programme information above).

Our results suggest that besides this ‘simple’ identification, the programme did not make any difference. In this respect, differences between the supported and rejected firms in the control group seem more random, rather than an indication of systematic programme impact. This claim is supported by the result from our analysis of the multiple-time supported firms. These firms were judged as being especially promising, and this judgement was correct, but only with respect to realization of sales during the first years. It seems that this in case too, that SIC were more inclined to support firms with average prospects. This behaviour is in line with what Wallsten (2000) proposed in his comment on the SBIR programme in the USA, that is, administrators are inclined to select firms that exhibit early results and which make the administrators look produc-tive. Unfortunately, few of the firms studied change from not being commercialized to being commercialized during the years that follow. Other probable explanations to this result could be that innovative ideas, compared to more ‘basic’ ideas are (1) difficult to communicate and evalu-ate and (2) are associated with great uncertainty (Wsthead and Storey, 1997; Klofsten, 2005; Norrman, 2008).

614 International Small Business Journal 28(6)

A general explanation for the lack of difference between supported and rejected firms might be that, given the small amounts allocated and the time and money invested in the selection, the results are highly influenced by randomness. For comparison, a small VC firm, Rendera,14 that invests €33–76 million in seed phases, can be mentioned. Rendera spends on average €5500 on due diligence for each case in which they invest (lower sums are spent on rejected firms). Still their failure rate, along with that of other actors in the VC branches, is substantial (Chrisman and McMullan, 2000; Zacharakis and Meyer, 2000; Lerner, 2002). Rendera calculates that 10–20% of their portfolio firms will grow enough to cover their total investment costs. From this comparison, we can conclude that the inability of SIC to pick winners could be coupled to the prerequisites for making decisions. The question is therefore, not merely whether this investigation of the applicants was too superficial, but also if it is at all possible to make a valid judgment at such early stages.

Second, although the aim of SIC was to enhance commercialization, most of the money lent was spent on product improvement or protection and only a minor part on marketing. This is regarded as a weakness according to a survey study of innovators’ attitudes within this programme (Norrman and Klofsten, 2009). In Norrman and Klofsten’s study, there were complaints about the lack of support tak-ing the idea to the market – a patent does not secure market success (Åstebro, 2003). Hence, this focus on product development and protection might be one explanation as to why additionality, in the case of economic performance (success in commercialization), as result of this programme, was not found.

Third, the time period studied (1995–2003) is, in general, recognized to be a period during which the supply of VC increased (EVCA, 2006) and thus, the probabilities of being able to raise funding on the private market could also have been an option for the firms that were rejected by SIC. Furthermore, the business cycle, with exception of the ICT drop, was strong, which created an advantageous economic environment. These circumstances imply that public programmes had to make a substantial impact for their efforts to emerge.

A final explanation is that the time span, eight years, might be too short, since it takes time for effects to emerge (Lundström and Stevenson, 2005). We indeed believe that a longer time series would give valuable information on firm dynamics. However, the possible impact of support pro-grammes becomes less certain the longer the time that passes since the development of the firm and is also affected by other important factors. In Sweden today, policies have moved towards the cre-ation of incubator environments, which implies that more resources are being put into selection procedures. The content of the programmes has also been more developed, that is, currently there is more focus on market and customer issues. If this development crowds out private venture capi-tal in early stages or actually invites it to earlier commitments is a topic for future research.

Concluding remarks and implicationsThe aim of this article was to investigate the impact of a policy programme directed to supporting early stage innovative ventures by making an attempt to identify the existence of the additionality created by the programme. We also aimed to contribute to the discussion about the measurement of the over time dynamics of early stages public support and to the debate on the evaluation of public support for such ventures. In order to reach these aims, we hypothesized that supported firms per-formed better than those firms which were not supported, and that firms that were multiple sup-ported performed better than those that only applied and gained support once. The hypotheses were tested on cohorts of firms annually, one to eight years after applications were filed.

Self-selection bias was taken into account since all the firms had applied for funding. By use of the matched pair estimation method of Abadie et al. (2004) we took into account the relatively small size of our quasi-control group, as well as the administrative selection bias. Our analysis is

Norrman and Bager-Sjögren 615

based on objective data, which according to law, is checked by external auditors. Thus, we regard our results as reliable and valid, and draw the conclusion that this study demonstrates that impact can be traced and measured from annual report data. Summarizing the results of our hypotheses, the following conclusions can be reached:

• The evidence of an impact of the support to early stages ventures given by the public programme SIC is weak or non-existent.

• The higher number of outliers in the supported groups could be an indication of prospective success if the time span is prolonged over seven years.

• Our test of the projects that programme officials considered to be most promising did not support their belief.

The fact that no additionality resulting from the programme has been detected is serious. However, it does not necessarily imply programme failure but instead that further research using a mixed methods approach is necessary to find explanations.

The second finding could be interpreted either as a random event (pure luck) or as an indication that the administrators actually were able to spot some innovators with high potential ideas. Unfortunately however, the performance of these has not been large enough to generate a positive overall impact of the programme in the time span studied. The third finding indicates the impact of randomness. Picking winners, especially in case of predicting what technique will be a future winner is difficult as it depends on a wide range of variables (Cusumano et al., 1992). Hence, our third find-ing suggests that, given eligible applications, supported firms could just as well be picked randomly. Furthermore, it could be that administrators striving to find winners instead tend towards the above-described Wallsten-effect that is, they tend to support more simple ideas which are better prospects in the short term (and thus give positive feedback on the lower level type of evaluations that Storey [2000] terms ‘monitoring’) gets supported, while radically innovative ideas are overlooked because they are high risk. Based on this we conclude that picking winners ought to be left to stages where firms are more developed, as we show that this requires both resources and information.

Finally we argue, along with Jaffe (2002) and Storey (2000), that it is of great importance that policymakers ensure that their programmes have developed evaluative awareness and in addition to this, define a clear counter-factual to measure against. Proper prerequisites for evaluation, where there are clearly stated measurable goals and indicators, are factors which also facilitate follow-ups. We also recommend that public support programmes are required to develop data collection strategies. As for further research it would be interesting to investigate whether this result also holds for firms that did not apply as limited firms, since the ideas of such ventures might be found in earlier development phases. It is therefore, possible that these ventures would have experienced a stronger development through this finance. Hence, we suggest that other legal liabilities, such as sole proprietors, are examined.

Notes

1. Almi is an organization supplying finance and advice to Swedish SME firms. 2. Swedepark has now reorganized and changed name into SISP, Swedish Incubators and Science Parks. 3. €1 = 9.197 SEK (5 year mean 30 June 2002–1 July 2007, Bank of Sweden). 4. The requirement of newness implies that, in most cases, SIC fundable firms suffered from the obstacles

of getting finance that are referred to section “public support and the scheme studied” and the subsection “the programme” at page 2.

616 International Small Business Journal 28(6)

5. Information taken from interviews with one or more of the following SIC officials: Per Laurell CEO 2002–2004, Åke Wallin CEO 1995–2001 and Roger Yttergren. The average rates of support were 57% in total and 66% for limited companies.

6. Annual reports from 1998, 1999, 2000, 2001, 2002, and two descriptions of the outcomes of the pro-gramme made by CMA, 2002 and 2003.

7. See Hoogerwerf (1990) and Hjalmarsson and Johansson (2003). 8. Later we substitute non-supported with ‘rejected’ in order to deal with self-selection “Data and estimation

method” and its subsection “sources of systematic errors” at p 7. 9. Abadie et al. (2004) presented an application for the STATA package (used in this analysis), that facili-

tates the estimation of the matching estimator. In their application, we utilize an option which allows us to use the method of bootstrapping in order to deal with the problem of a small control group. Not using this technique would have led to bias in the estimation of the variance, that is, the variance would have been underestimated.

10. In the language of statistical hypothesis testing, our hypotheses are the so-called alternative hypotheses. They are tested against the H-naught of: Difference between supported and rejected firms being equal or less than zero. Our statistics are thus significant if this H-naught is rejected (and in our language, such a rejection is a support to our hypothesis).

11. The most popular matching estimator is the difference-in-difference, or fixed effect estimator, which our application of matching variables resembles. However, this hinges upon certain assumptions regarding the equality between treated and non-treated groups in the pre-programme period (Heckman and Robb, 1985: 216) which we have in part secured for.

12. In order to further describe the outliers, see table below where kurtosis and the sum of the two largest values is presented

13. Tables withheld for space reasons, available from the authors.14. Rendera AB, Interview with Björn Persson, Investment Manager.

AcknowledgementsWe are grateful to Jonas Månsson, Maria Minniti and the anonymous referees for their productive sugges-tions. Thanks also to our colleges, the former personnel of SIC, and the helpful people engaged in the associa-tions of inventors for productive comments on our work.

Kurtosis Sum of two largest items (in milj sek)

Year Rejected Single-app M-appli Rejected Single-app M-appli

1 17 290 14 8.3 53 4.22 40 324 12 12.5 74.5 5.13 29 286 24 11.6 73 7.54 16 176 52 8.2 106 11.35 15 154 48 11.7 106 136 8 97 43 10.4 117 13.27 8 83 41 10.3 114 13.3

Year=number of years after application were made, Rejected = rejected applications, Single-app =supported firms that have applied for funding once, M-appli= supported firms that have applied more than one time and never have been rejected.

Table 6. The statistics “kurtosis” and “sum of two largest values” accentuate the fat tails in the distribution of sales for single-applicants.

Norrman and Bager-Sjögren 617

ReferencesAbadie A and Imbens G (2002) Simple and bias-corrected matching estimators for average treatment effects.

NBER Technical Working Paper 283.Abadie A, Drukker D, Herr JL and Imbens G (2004) Implementing matching estimators for average treatment

effects in stata. Stata Journal 4(3): 290–311.Almus M (2004) Job creation through public start-up assistance? Applied Economics 36(18): 2015–2024.Åstebro T (2003) The return to independent invention: Evidence of unrealistic optimism, risk seeking or

skewness loving? The Economic Journal 113: 226–239.Audretsch DB (2002) Entrepreneurship: A Survey of the Literature. European Commission, Enterprise

Directorate General.Bager-Sjögren L and Lööf H (2005) The Impact on Growth from Public Seed Financing to New Technology

Projects in Small Enterprises. Stockholm: ITPS.Barney J (1991) Firm resources and sustained competitive advantage. Journal of Management 17(1): 99.Baum J and Silverman B (2004) Picking winners or building them? Alliance, intellectual, and human capital

as selection criteria in venture financing and performance of biotechnology start-ups. Journal of Business Venturing 19: 411–436.

Bygrave WD and Timmons JA (1992) Venture Capital at the Crossroads. Boston, MA: Harvard Business School Press.

Chrisman JJ and McMullan EW (2000) A preliminary assessment of outsider assistance as a knowledge resource: The longer-term impact of new venture counselling. Entrepreneurship: Theory & Practice 24(3): 37.

COM (2005) Decision of the European Parliament and the Council.COM (2006) Communication from the Commission to the Council, the European Parliament, the European

Economic and Social Committee and the Committee of the Regions.Cusumano MA, Mylonadis Y and Rosenbloom RS (1992) Strategic maneuvering and mass-market dynamics:

The triumph of VHS over Beta. Business History Review 66(1): 51.Davidsson P and Henrekson M (2002) Determinants of the prevalence of start-ups and high-growth firms.

Small Business Economics 19: 81–104.Davidsson P, Kirchhoff B, Hetemi-JA and Gustafsson H (2002) Empirical analysis of business growth factors

using Swedish data. Journal of Small Business Management 40: 332–349.Deakins D (1994) Risk assessment with asymmetric information. International Journal of Bank Marketing

Emerald 12: 24.Drucker, PF (1985) Innovation and Entrepreneurship. London: Heineman.Edling J, Hermansson K, Nilsson R and Nordborg J (2007) Innovativa Små Och Medelstora Företag –

Sveriges Framtid. VINNOVA policy VP 2007:01. Stockholm, VINNOVA: 96.EVCA (2006) Evca Yearbook 2006. Belgium: EVCA.Governmental Decision (1994) Regeringsbeslut 23: Upprättande Av Stiftelsen Innovationscentrum, 1994:862/94

(Governmental Decision on Inception of Sweden Innovation Centre).Governmental Proposition (1993/1994) Regeringens Proposition: Innovationsstöd, Prop. 1993/94:206.

(Governmental Proposition: Innovation Support).Harhoff D and Stahl K (1998) Legal form, growth and exit of West German firms – empirical results for

manufacturing, construction, trade and service industries. Journal of Industrial Economics 46: 453–488.Heckam J and Robb R (1985) Alternative methods for evaluating the impact of interventions. In: Heckman J

and Singer B (eds) Longitudinal Analysis of Labor Market Data. New York: Wiley.Heckman J, Ichimura H and Todd P (1997) Matching as an econometric evaluation estimator: Evidence from

evaluating a job training program. Review of Economic Studies 64(4): 605–654.Heydebreck P, Klofsten M and Maier J C (2000) Innovation support for new technology-based firms: The

Swedish Teknopol approach. R&D Management 30(1): 89–100.

618 International Small Business Journal 28(6)

Hjalmarsson D and Johansson AW (2003) Public advisory services – theory and practice. Entrepreneurship & Regional Development 15(1): 83.

Hoogerwerf A (1990) Reconstructing policy theory. Evaluation and Program Planning 13(3): 285–291.Jaffe AB (2002) Building programme evaluation into the design of public research-support programmes.

Oxford Review of Economic Policy 18(1): 22–34.Klofsten M (1992) Tidiga Utvecklingsprocesser i Teknikbaserade Företag (Early development processes in

technology-based firms). Dissertation, Linköping University.Klofsten M (2005) New venture ideas: An analysis of their origin and early development. Technology Analysis

& Strategic Management 17: 105–119.Klofsten M, Jonsson M and Simón J (1999) Supporting the pre-commercialization stages of technology-based

firms: The effects of small-scale venture capital. Venture Capital 1(1): 83–93.Leleux B and Surlemont B (2003) Public versus private venture capital: Seeding or crowding out? A Pan-

European analysis. Journal of Business Venturing 18: 81–104.Lerner J (2002) When bureaucrats meet entrepreneurs: The design of effective public venture capital pro-

grammes. Economic Journal 112: F73.Lundström A and Stevenson L (2002) On the Road to Entrepreneurship Policy. Stockholm: FSF.Lundström A and Stevenson L (2005) Entrepreneurship Policy: Theory and Practice. New York: Springer.Mainprize B and Hindle K (2007) The benefit: a well-written entrepreneurial business plan is to an entrepre-

neur what a midwife is to an expecting mother. Journal of Private Equity 11(1): 40–52.Mason C and Harrisson R (2002) Barriers to investment in the informal venture capital sector. Entrepreneurship

& Regional Development 14: 271–287.Meyer M (2005) Independent inventors and public support measures: Insights from 33 case studies in Finland.

World Patent Information 27: 113–123.Ministry of Trade and Industry (2006) En Nationell Strategi för Regional Konkurrenskraft, Entreprenörskap

Och Sysselsättning 2007–2013. (A National Strategy for Regional Competitiveness), Stockholm.Mosselman M, Prince Y and Kemp R (2004) Review of the methodologies to measure effectiveness of state

aid to SMEs. Final Report to the European Commission.Nagel SS (2002) Handbook of Public Policy Evaluation. Thousand Oaks, CA: SAGE.Norrman C (2005) Publicly funded support of technology-based ventures. Linköping Studies in Science and

Technology, Licentiate Thesis, No 1219, LiU, Tek-Lic 2005:72, Linköping University, Linköping.Norrman C (2008) Entrepreneurship policy: Public support for technology-based ventures. Dissertation

No.1175, Linköping Institute of Technology.Norrman C and Klofsten M (2009) An entrepreneurship policy programme: Implications and expectations.

International Journal of Entrepreneurship & Innovation 10(1): 33–42.North D, Smallbone D and Vickers I (2001) Public sector support for innovating SMEs. Small Business

Economics 16: 303–317.Oakey RP (2003) Funding innovation and growth in UK new technology-based firms: Some observations on

contributions from the public and private sectors. Venture Capital 5(2): 161–179.OECD (2006) Going for Growth. Paris: OECD.Penrose ET (1959) The Theory of the Growth of the Firm. London: Basil Blackwell.Pleiborn M (2002) Projekt Vid Stiftelsen Innovationscentrum (Projects at Sweden Innovation Centre)

Stockholm: Stiftelsen Innovationscentrum and Temaplan.Reitberger G (1983) STU-Projekt med Villkorlig Återbetalningsskyldighet – Analys och Utvärdering

av Effekter (STU-projects with conditional duty of repayment – analysis and evaluation of effects). Konsultrapport, Dnr 834014, September 1983, Styrelsen för Teknisk Utveckling.

Rosenbaum P and Rubin D (1984) Reducing the bias in observation studies using sub-classification on the propensity score. Journal of the American Statistical Association 79: 516–524.

Norrman and Bager-Sjögren 619

Rothschild M and Stiglitz J (1976) Equilibrium in competitive insurance markets: An essay on the economics of imperfect information. Quarterly Journal of Economics 90: 629–649.

Rush H, Bessant J and Lees S (2004) Assessing the effectiveness of technology policy – a long-term view. Technology Analysis & Strategic Management 16(3): 327–342.

Salmenkaita J-P and Salo A (2002) Rationales for government intervention in the commercialization of new technologies. Technology Analysis and Strategic Management 14(2): 183–200.

Shane S and Cable D (2002) Network ties, reputation, and the financing of new ventures. Management Science, INFORMS: Institute for Operations Research 48: 364.

SIC (1999) Årsredovisning. Stockholm: Sweden Innovation Centre.SIC (2002) Stiftelsen Innovationscentrum 2002 (Sweden Innovation Centre 2002). Stockholm: Stiftelsen

Innovationscentrum.SIC (2004) 10 År Med Stiftelsen Innovationscentrum (10 Years with Sweden Innovation Centre). Stockholm:

Stiftelsen Innovationscentrum: 1–15.Stam E, Suddle K, Hessles J and van Stel A (2007) High growth entrepreneurs, public policies and economic

growth. Jena Economic Research Papers 2007–2019.Stinchcombe AL (1965) Social Structure and Organizations. Handbook of Organizations. New York: Garland

Publishing.Storey DJ (1994) Understanding the Small Business Sector. London: Routledge.Storey DJ (2000) Six steps to heaven: Evaluating the impact of public policies to support small business in

developed economies. In: Sexton D and Landström H (eds) The Blackwell Handbook of Entrepreneurship. Oxford: Wiley-Blackwell, 176–193.

Storey D J (2003) Entrepreneurship: Small and Medium Sized Enterprises and Public Policies. Handbook of Entrepreneurship Research. Amsterdam: Kluwer.

Storey DJ and Tether BS (1998) New technology-based firms in the European Union: An introduction. Research Policy 26(9): 933–946.

Svensson R (2007) Commercialization of Patents and External Financing during the R&D-Phase, Research Policy, Vol. 36(7), s. 1052–69, 2007.

Wallsten SJ (2000) The effects of government-industry R&D programs on private R&D: The case of the small business innovation research program. RAND Journal of Economics 31(1): 82–100.

Vedin BA (1993) Innovationer för Sverige: Betänkande Av Innovationsutredningen (Innovations for Sweden), SOU 1993:84, Näringsdepartementet, Stockholm.

Vedung E (1998) Utvärdering I Politik och Förvaltning (Evaluation in Politics and Administration). Lund: Studentlitteratur.

Westhead P and Storey DJ (1997) Financial constraints on the growth of high technology small firms in the United Kingdom. Applied Financial Economics 7(2): 197–201.

Zacharakis AL and Meyer GD (2000) The potential of actuarial decision models: Can they improve the ven-ture capital investment decision? Journal of Business Venturing 15: 323–346.

Charlotte Norrman, PhD, is assistant professor at Linköping University. She teaches entrepreneurship and inno-vation courses. Her main research interest is entrepreneurship policy issues and early stage technology-based ventures. Before her doctoral studies (2003–2008), she worked several years in organizations that supported entrepreneurship, most recently (1998–2003), as a member of the founder team of Västerås Technology Park.

Lars Bager-Sjögren is senior economist at the Swedish Agency for Growth Policy Analysis (former Institute for growth policies, ITPS). He holds a licentiate degree in economics. Lars has made several evaluations of policy initiatives at ITPS; he has also participated in several research projects.